@detiq/crawler 1.0.6 → 1.0.8

package/dist/extractors/advanced.js

@@ -83,11 +83,18 @@ function extractMetadata(html, url) {
     ogImage: null,
     ogUrl: null,
     ogType: null,
+    ogSiteName: null,
+    ogLocale: null,
     twitterCard: null,
     twitterTitle: null,
     twitterDescription: null,
+    twitterImage: null,
+    twitterSite: null,
     canonical: null,
     robots: null,
+    hreflang: [],
+    themeColor: null,
+    favicon: null,
     jsonLd: [],
     schemaTypes: [],
     lang: null
@@ -107,14 +114,27 @@ function extractMetadata(html, url) {
       if (name === "twitter:card") meta.twitterCard = content;
       if (name === "twitter:title") meta.twitterTitle = content;
       if (name === "twitter:description") meta.twitterDescription = content;
+      if (name === "twitter:image") meta.twitterImage = content;
+      if (name === "twitter:site") meta.twitterSite = content;
+      if (name === "theme-color") meta.themeColor = content;
       if (prop === "og:title") meta.ogTitle = content;
       if (prop === "og:description") meta.ogDescription = content;
       if (prop === "og:image") meta.ogImage = content;
       if (prop === "og:url") meta.ogUrl = content;
       if (prop === "og:type") meta.ogType = content;
+      if (prop === "og:site_name") meta.ogSiteName = content;
+      if (prop === "og:locale") meta.ogLocale = content;
     }
     const canonical = document.querySelector('link[rel="canonical"]');
     if (canonical) meta.canonical = canonical.getAttribute("href") ?? null;
+    const faviconEl = document.querySelector('link[rel="icon"], link[rel="shortcut icon"]');
+    if (faviconEl) meta.favicon = faviconEl.getAttribute("href") ?? null;
+    const hreflangEls = [...document.querySelectorAll('link[rel="alternate"][hreflang]')];
+    for (const el of hreflangEls) {
+      const lang = el.getAttribute("hreflang") ?? "";
+      const href = el.getAttribute("href") ?? "";
+      if (lang && href) meta.hreflang.push({ lang, href });
+    }
     const jsonLdScripts = [...document.querySelectorAll('script[type="application/ld+json"]')];
     for (const s of jsonLdScripts) {
       try {

package/dist/extractors/advanced.js.map

@@ -1,7 +1,7 @@
 {
   "version": 3,
   "sources": ["../../src/extractors/advanced.ts"],
-  "sourcesContent": ["import { parseHTML } from 'linkedom';\n\nexport interface ExtractionResult {\n  matches: string[];\n  count: number;\n  method: string;\n}\n\nexport type TableRow = Record<string, string>;\n\nexport interface PageMetadata {\n  title: string | null;\n  description: string | null;\n  keywords: string | null;\n  ogTitle: string | null;\n  ogDescription: string | null;\n  ogImage: string | null;\n  ogUrl: string | null;\n  ogType: string | null;\n  twitterCard: string | null;\n  twitterTitle: string | null;\n  twitterDescription: string | null;\n  canonical: string | null;\n  robots: string | null;\n  jsonLd: object[];\n  schemaTypes: string[];\n  lang: string | null;\n}\n\nexport function extractByCssSelector(html: string, selector: string, attr?: string): ExtractionResult {\n  try {\n    const { document } = parseHTML(html);\n    const nodes = Array.from(document.querySelectorAll(selector));\n    const matches = nodes.map((n: any) => {\n      if (attr) return (n as any).getAttribute(attr) ?? '';\n      return (n as any).textContent?.trim() ?? '';\n    }).filter(Boolean);\n    return { matches, count: matches.length, method: 'css-selector' };\n  } catch (e) {\n    return { matches: [], count: 0, method: 'css-selector-error' };\n  }\n}\n\nexport function extractByXPath(html: string, xpath: string): ExtractionResult {\n  try {\n    const { document, XPathResult } = parseHTML(html) as any;\n    const result = document.evaluate(xpath, document, null, XPathResult?.ANY_TYPE ?? 0, null);\n    const matches: string[] = [];\n    let node = result.iterateNext?.();\n    while (node) {\n      const text = (node as any).textContent?.trim() ?? (node as any).nodeValue?.trim() ?? '';\n      if (text) matches.push(text);\n      node = result.iterateNext?.();\n    }\n    return { matches, count: matches.length, method: 'xpath' };\n  } catch (e) {\n    return { matches: [], count: 0, method: 'xpath-error' };\n  }\n}\n\nexport function extractByRegex(text: string, pattern: string, flags?: string): ExtractionResult {\n  try {\n    const re = new RegExp(pattern, (flags ?? 'g').includes('g') ? flags : (flags ?? '') + 'g');\n    const matches: string[] = [];\n    let m: RegExpExecArray | null;\n    while ((m = re.exec(text)) !== null) {\n      matches.push(m[1] ?? m[0]);\n      if (!re.global) break;\n    }\n    return { matches, count: matches.length, method: 'regex' };\n  } catch {\n    return { matches: [], count: 0, method: 'regex-error' };\n  }\n}\n\nexport function extractTables(html: string): TableRow[][] {\n  try {\n    const { document } = parseHTML(html);\n    const tables = Array.from(document.querySelectorAll('table'));\n    return tables.map((table: any) => {\n      const headers: string[] = [];\n      const rows: TableRow[] = [];\n      const ths = [...table.querySelectorAll('thead th, thead td, tr:first-child th')];\n      if (ths.length > 0) {\n        ths.forEach((th: any, i: number) => headers.push(th.textContent?.trim() || `col${i + 1}`));\n      }\n      const trs = [...table.querySelectorAll('tbody tr, tr')];\n      for (const tr of trs) {\n        const cells = [...(tr as any).querySelectorAll('td, th')];\n        if (cells.length === 0) continue;\n        if (headers.length === 0) {\n          cells.forEach((_: any, i: number) => headers.push(`col${i + 1}`));\n        }\n        const row: TableRow = {};\n        cells.forEach((cell: any, i: number) => {\n          const key = headers[i] || `col${i + 1}`;\n          row[key] = cell.textContent?.trim() ?? '';\n        });\n        if (Object.values(row).some(v => v)) rows.push(row);\n      }\n      return rows;\n    }).filter(t => t.length > 0);\n  } catch {\n    return [];\n  }\n}\n\nexport function extractMetadata(html: string, url: string): PageMetadata {\n  const meta: PageMetadata = {\n    title: null, description: null, keywords: null,\n    ogTitle: null, ogDescription: null, ogImage: null, ogUrl: null, ogType: null,\n    twitterCard: null, twitterTitle: null, twitterDescription: null,\n    canonical: null, robots: null, jsonLd: [], schemaTypes: [], lang: null,\n  };\n  try {\n    const { document } = parseHTML(html);\n    meta.title = (document as any).title ?? (document as any).querySelector('title')?.textContent?.trim() ?? null;\n    meta.lang = (document as any).documentElement?.getAttribute('lang') ?? null;\n\n    const metas = [...(document as any).querySelectorAll('meta')];\n    for (const m of metas) {\n      const name = (m as any).getAttribute('name')?.toLowerCase() ?? '';\n      const prop = (m as any).getAttribute('property')?.toLowerCase() ?? '';\n      const content = (m as any).getAttribute('content') ?? '';\n      if (name === 'description') meta.description = content;\n      if (name === 'keywords') meta.keywords = content;\n      if (name === 'robots') meta.robots = content;\n      if (name === 'twitter:card') meta.twitterCard = content;\n      if (name === 'twitter:title') meta.twitterTitle = content;\n      if (name === 'twitter:description') meta.twitterDescription = content;\n      if (prop === 'og:title') meta.ogTitle = content;\n      if (prop === 'og:description') meta.ogDescription = content;\n      if (prop === 'og:image') meta.ogImage = content;\n      if (prop === 'og:url') meta.ogUrl = content;\n      if (prop === 'og:type') meta.ogType = content;\n    }\n\n    const canonical = (document as any).querySelector('link[rel=\"canonical\"]');\n    if (canonical) meta.canonical = canonical.getAttribute('href') ?? null;\n\n    const jsonLdScripts = [...(document as any).querySelectorAll('script[type=\"application/ld+json\"]')];\n    for (const s of jsonLdScripts) {\n      try {\n        const parsed = JSON.parse((s as any).textContent ?? '{}');\n        meta.jsonLd.push(parsed);\n        const types = Array.isArray(parsed) ? parsed.map((p: any) => p['@type']).flat() : [parsed['@type']];\n        meta.schemaTypes.push(...types.filter(Boolean));\n      } catch { /* skip invalid JSON-LD */ }\n    }\n  } catch { /* non-fatal */ }\n  return meta;\n}\n\nexport interface ProductData {\n  name: string | null;\n  description: string | null;\n  brand: string | null;\n  sku: string | null;\n  price: number | null;\n  priceCurrency: string | null;\n  availability: string | null;\n  rating: number | null;\n  reviewCount: number | null;\n  image: string | null;\n  url: string | null;\n}\n\nexport function extractStructuredProduct(jsonLd: object[]): ProductData | null {\n  for (const item of jsonLd) {\n    try {\n      const type = (item as any)['@type'];\n      const isProduct = Array.isArray(type)\n        ? type.some((t: any) => typeof t === 'string' && t.toLowerCase() === 'product')\n        : typeof type === 'string' && type.toLowerCase() === 'product';\n      if (!isProduct) continue;\n\n      const i = item as any;\n      const offers = Array.isArray(i.offers) ? i.offers[0] : i.offers;\n\n      const rawPrice = offers?.price;\n      const price = rawPrice != null ? parseFloat(rawPrice) : NaN;\n\n      const rawRating = i.aggregateRating?.ratingValue;\n      const rating = rawRating != null ? parseFloat(rawRating) : NaN;\n\n      const rawReviewCount = i.aggregateRating?.reviewCount;\n      const reviewCount = rawReviewCount != null ? parseInt(rawReviewCount, 10) : NaN;\n\n      const rawAvailability: string | undefined = offers?.availability;\n      const availability = rawAvailability\n        ? rawAvailability.replace(/^https?:\\/\\/schema\\.org\\//, '')\n        : null;\n\n      const rawImage = i.image;\n      const image = Array.isArray(rawImage) ? (rawImage[0] ?? null) : (rawImage ?? null);\n\n      const brand = i.brand?.name ?? i.brand ?? null;\n\n      return {\n        name: i.name ?? null,\n        description: i.description ?? null,\n        brand: typeof brand === 'string' ? brand : null,\n        sku: i.sku ?? null,\n        price: isNaN(price) ? null : price,\n        priceCurrency: offers?.priceCurrency ?? null,\n        availability,\n        rating: isNaN(rating) ? null : rating,\n        reviewCount: isNaN(reviewCount) ? null : reviewCount,\n        image,\n        url: i.url ?? null,\n      };\n    } catch { /* skip malformed item */ }\n  }\n  return null;\n}\n\nexport function extractImages(html: string, baseUrl: string): Array<{ src: string; alt: string; width?: number; height?: number }> {\n  try {\n    const { document } = parseHTML(html);\n    const base = new URL(baseUrl);\n    return Array.from((document as any).querySelectorAll('img'))\n      .map((img: any) => {\n        const src = img.getAttribute('src') ?? img.getAttribute('data-src') ?? '';\n        if (!src) return null;\n        try {\n          return {\n            src: new URL(src, base).href,\n            alt: (img.getAttribute('alt') ?? '') as string,\n            width: img.getAttribute('width') ? parseInt(img.getAttribute('width')) : undefined,\n            height: img.getAttribute('height') ? parseInt(img.getAttribute('height')) : undefined,\n          } as { src: string; alt: string; width?: number; height?: number };\n        } catch { return null; }\n      })\n      .filter((x): x is { src: string; alt: string; width?: number; height?: number } => x !== null);\n  } catch {\n    return [];\n  }\n}\n\nexport function extractMediaFiles(html: string, baseUrl: string): Array<{ url: string; type: 'image' | 'video' | 'audio' | 'document'; tag: string }> {\n  const IMAGE_EXT = /\\.(png|jpe?g|gif|webp|svg|ico|avif)(\\?.*)?$/i;\n  const VIDEO_EXT = /\\.(mp4|webm|ogv|mov|avi)(\\?.*)?$/i;\n  const AUDIO_EXT = /\\.(mp3|ogg|wav|flac|aac)(\\?.*)?$/i;\n  const DOC_EXT = /\\.(pdf|docx?|xlsx?|pptx?|csv|zip)(\\?.*)?$/i;\n\n  function classify(url: string, tag: string): 'image' | 'video' | 'audio' | 'document' {\n    if (tag === 'img' || IMAGE_EXT.test(url)) return 'image';\n    if (tag === 'video' || VIDEO_EXT.test(url)) return 'video';\n    if (tag === 'audio' || AUDIO_EXT.test(url)) return 'audio';\n    if (DOC_EXT.test(url)) return 'document';\n    return 'document';\n  }\n\n  try {\n    const { document } = parseHTML(html);\n    const base = new URL(baseUrl);\n    const results: Array<{ url: string; type: 'image' | 'video' | 'audio' | 'document'; tag: string }> = [];\n    const seen = new Set<string>();\n\n    const selectors: [string, string] = ['img[src],video[src],audio[src],source[src]', 'src'];\n    const attrPairs: Array<[string, string]> = [\n      ['img', 'src'], ['video', 'src'], ['audio', 'src'], ['source', 'src'],\n      ['a', 'href'],\n    ];\n\n    for (const [sel, attr] of attrPairs) {\n      for (const el of [...(document as any).querySelectorAll(sel)]) {\n        const raw = (el as any).getAttribute(attr) ?? '';\n        if (!raw) continue;\n        try {\n          const resolved = new URL(raw, base).href;\n          if (seen.has(resolved)) continue;\n          if (sel === 'a' && !DOC_EXT.test(resolved) && !IMAGE_EXT.test(resolved) && !VIDEO_EXT.test(resolved) && !AUDIO_EXT.test(resolved)) continue;\n          seen.add(resolved);\n          results.push({ url: resolved, type: classify(resolved, sel), tag: sel });\n        } catch { /* skip */ }\n      }\n    }\n    return results;\n  } catch {\n    return [];\n  }\n}\n"],
-  "mappings": "AAAA,SAAS,iBAAiB;AA6BnB,SAAS,qBAAqB,MAAc,UAAkB,MAAiC;AACpG,MAAI;AACF,UAAM,EAAE,SAAS,IAAI,UAAU,IAAI;AACnC,UAAM,QAAQ,MAAM,KAAK,SAAS,iBAAiB,QAAQ,CAAC;AAC5D,UAAM,UAAU,MAAM,IAAI,CAAC,MAAW;AACpC,UAAI,KAAM,QAAQ,EAAU,aAAa,IAAI,KAAK;AAClD,aAAQ,EAAU,aAAa,KAAK,KAAK;AAAA,IAC3C,CAAC,EAAE,OAAO,OAAO;AACjB,WAAO,EAAE,SAAS,OAAO,QAAQ,QAAQ,QAAQ,eAAe;AAAA,EAClE,SAAS,GAAG;AACV,WAAO,EAAE,SAAS,CAAC,GAAG,OAAO,GAAG,QAAQ,qBAAqB;AAAA,EAC/D;AACF;AAEO,SAAS,eAAe,MAAc,OAAiC;AAC5E,MAAI;AACF,UAAM,EAAE,UAAU,YAAY,IAAI,UAAU,IAAI;AAChD,UAAM,SAAS,SAAS,SAAS,OAAO,UAAU,MAAM,aAAa,YAAY,GAAG,IAAI;AACxF,UAAM,UAAoB,CAAC;AAC3B,QAAI,OAAO,OAAO,cAAc;AAChC,WAAO,MAAM;AACX,YAAM,OAAQ,KAAa,aAAa,KAAK,KAAM,KAAa,WAAW,KAAK,KAAK;AACrF,UAAI,KAAM,SAAQ,KAAK,IAAI;AAC3B,aAAO,OAAO,cAAc;AAAA,IAC9B;AACA,WAAO,EAAE,SAAS,OAAO,QAAQ,QAAQ,QAAQ,QAAQ;AAAA,EAC3D,SAAS,GAAG;AACV,WAAO,EAAE,SAAS,CAAC,GAAG,OAAO,GAAG,QAAQ,cAAc;AAAA,EACxD;AACF;AAEO,SAAS,eAAe,MAAc,SAAiB,OAAkC;AAC9F,MAAI;AACF,UAAM,KAAK,IAAI,OAAO,UAAU,SAAS,KAAK,SAAS,GAAG,IAAI,SAAS,SAAS,MAAM,GAAG;AACzF,UAAM,UAAoB,CAAC;AAC3B,QAAI;AACJ,YAAQ,IAAI,GAAG,KAAK,IAAI,OAAO,MAAM;AACnC,cAAQ,KAAK,EAAE,CAAC,KAAK,EAAE,CAAC,CAAC;AACzB,UAAI,CAAC,GAAG,OAAQ;AAAA,IAClB;AACA,WAAO,EAAE,SAAS,OAAO,QAAQ,QAAQ,QAAQ,QAAQ;AAAA,EAC3D,QAAQ;AACN,WAAO,EAAE,SAAS,CAAC,GAAG,OAAO,GAAG,QAAQ,cAAc;AAAA,EACxD;AACF;AAEO,SAAS,cAAc,MAA4B;AACxD,MAAI;AACF,UAAM,EAAE,SAAS,IAAI,UAAU,IAAI;AACnC,UAAM,SAAS,MAAM,KAAK,SAAS,iBAAiB,OAAO,CAAC;AAC5D,WAAO,OAAO,IAAI,CAAC,UAAe;AAChC,YAAM,UAAoB,CAAC;AAC3B,YAAM,OAAmB,CAAC;AAC1B,YAAM,MAAM,CAAC,GAAG,MAAM,iBAAiB,uCAAuC,CAAC;AAC/E,UAAI,IAAI,SAAS,GAAG;AAClB,YAAI,QAAQ,CAAC,IAAS,MAAc,QAAQ,KAAK,GAAG,aAAa,KAAK,KAAK,MAAM,IAAI,CAAC,EAAE,CAAC;AAAA,MAC3F;AACA,YAAM,MAAM,CAAC,GAAG,MAAM,iBAAiB,cAAc,CAAC;AACtD,iBAAW,MAAM,KAAK;AACpB,cAAM,QAAQ,CAAC,GAAI,GAAW,iBAAiB,QAAQ,CAAC;AACxD,YAAI,MAAM,WAAW,EAAG;AACxB,YAAI,QAAQ,WAAW,GAAG;AACxB,gBAAM,QAAQ,CAAC,GAAQ,MAAc,QAAQ,KAAK,MAAM,IAAI,CAAC,EAAE,CAAC;AAAA,QAClE;AACA,cAAM,MAAgB,CAAC;AACvB,cAAM,QAAQ,CAAC,MAAW,MAAc;AACtC,gBAAM,MAAM,QAAQ,CAAC,KAAK,MAAM,IAAI,CAAC;AACrC,cAAI,GAAG,IAAI,KAAK,aAAa,KAAK,KAAK;AAAA,QACzC,CAAC;AACD,YAAI,OAAO,OAAO,GAAG,EAAE,KAAK,OAAK,CAAC,EAAG,MAAK,KAAK,GAAG;AAAA,MACpD;AACA,aAAO;AAAA,IACT,CAAC,EAAE,OAAO,OAAK,EAAE,SAAS,CAAC;AAAA,EAC7B,QAAQ;AACN,WAAO,CAAC;AAAA,EACV;AACF;AAEO,SAAS,gBAAgB,MAAc,KAA2B;AACvE,QAAM,OAAqB;AAAA,IACzB,OAAO;AAAA,IAAM,aAAa;AAAA,IAAM,UAAU;AAAA,IAC1C,SAAS;AAAA,IAAM,eAAe;AAAA,IAAM,SAAS;AAAA,IAAM,OAAO;AAAA,IAAM,QAAQ;AAAA,IACxE,aAAa;AAAA,IAAM,cAAc;AAAA,IAAM,oBAAoB;AAAA,IAC3D,WAAW;AAAA,IAAM,QAAQ;AAAA,IAAM,QAAQ,CAAC;AAAA,IAAG,aAAa,CAAC;AAAA,IAAG,MAAM;AAAA,EACpE;AACA,MAAI;AACF,UAAM,EAAE,SAAS,IAAI,UAAU,IAAI;AACnC,SAAK,QAAS,SAAiB,SAAU,SAAiB,cAAc,OAAO,GAAG,aAAa,KAAK,KAAK;AACzG,SAAK,OAAQ,SAAiB,iBAAiB,aAAa,MAAM,KAAK;AAEvE,UAAM,QAAQ,CAAC,GAAI,SAAiB,iBAAiB,MAAM,CAAC;AAC5D,eAAW,KAAK,OAAO;AACrB,YAAM,OAAQ,EAAU,aAAa,MAAM,GAAG,YAAY,KAAK;AAC/D,YAAM,OAAQ,EAAU,aAAa,UAAU,GAAG,YAAY,KAAK;AACnE,YAAM,UAAW,EAAU,aAAa,SAAS,KAAK;AACtD,UAAI,SAAS,cAAe,MAAK,cAAc;AAC/C,UAAI,SAAS,WAAY,MAAK,WAAW;AACzC,UAAI,SAAS,SAAU,MAAK,SAAS;AACrC,UAAI,SAAS,eAAgB,MAAK,cAAc;AAChD,UAAI,SAAS,gBAAiB,MAAK,eAAe;AAClD,UAAI,SAAS,sBAAuB,MAAK,qBAAqB;AAC9D,UAAI,SAAS,WAAY,MAAK,UAAU;AACxC,UAAI,SAAS,iBAAkB,MAAK,gBAAgB;AACpD,UAAI,SAAS,WAAY,MAAK,UAAU;AACxC,UAAI,SAAS,SAAU,MAAK,QAAQ;AACpC,UAAI,SAAS,UAAW,MAAK,SAAS;AAAA,IACxC;AAEA,UAAM,YAAa,SAAiB,cAAc,uBAAuB;AACzE,QAAI,UAAW,MAAK,YAAY,UAAU,aAAa,MAAM,KAAK;AAElE,UAAM,gBAAgB,CAAC,GAAI,SAAiB,iBAAiB,oCAAoC,CAAC;AAClG,eAAW,KAAK,eAAe;AAC7B,UAAI;AACF,cAAM,SAAS,KAAK,MAAO,EAAU,eAAe,IAAI;AACxD,aAAK,OAAO,KAAK,MAAM;AACvB,cAAM,QAAQ,MAAM,QAAQ,MAAM,IAAI,OAAO,IAAI,CAAC,MAAW,EAAE,OAAO,CAAC,EAAE,KAAK,IAAI,CAAC,OAAO,OAAO,CAAC;AAClG,aAAK,YAAY,KAAK,GAAG,MAAM,OAAO,OAAO,CAAC;AAAA,MAChD,QAAQ;AAAA,MAA6B;AAAA,IACvC;AAAA,EACF,QAAQ;AAAA,EAAkB;AAC1B,SAAO;AACT;AAgBO,SAAS,yBAAyB,QAAsC;AAC7E,aAAW,QAAQ,QAAQ;AACzB,QAAI;AACF,YAAM,OAAQ,KAAa,OAAO;AAClC,YAAM,YAAY,MAAM,QAAQ,IAAI,IAChC,KAAK,KAAK,CAAC,MAAW,OAAO,MAAM,YAAY,EAAE,YAAY,MAAM,SAAS,IAC5E,OAAO,SAAS,YAAY,KAAK,YAAY,MAAM;AACvD,UAAI,CAAC,UAAW;AAEhB,YAAM,IAAI;AACV,YAAM,SAAS,MAAM,QAAQ,EAAE,MAAM,IAAI,EAAE,OAAO,CAAC,IAAI,EAAE;AAEzD,YAAM,WAAW,QAAQ;AACzB,YAAM,QAAQ,YAAY,OAAO,WAAW,QAAQ,IAAI;AAExD,YAAM,YAAY,EAAE,iBAAiB;AACrC,YAAM,SAAS,aAAa,OAAO,WAAW,SAAS,IAAI;AAE3D,YAAM,iBAAiB,EAAE,iBAAiB;AAC1C,YAAM,cAAc,kBAAkB,OAAO,SAAS,gBAAgB,EAAE,IAAI;AAE5E,YAAM,kBAAsC,QAAQ;AACpD,YAAM,eAAe,kBACjB,gBAAgB,QAAQ,6BAA6B,EAAE,IACvD;AAEJ,YAAM,WAAW,EAAE;AACnB,YAAM,QAAQ,MAAM,QAAQ,QAAQ,IAAK,SAAS,CAAC,KAAK,OAAS,YAAY;AAE7E,YAAM,QAAQ,EAAE,OAAO,QAAQ,EAAE,SAAS;AAE1C,aAAO;AAAA,QACL,MAAM,EAAE,QAAQ;AAAA,QAChB,aAAa,EAAE,eAAe;AAAA,QAC9B,OAAO,OAAO,UAAU,WAAW,QAAQ;AAAA,QAC3C,KAAK,EAAE,OAAO;AAAA,QACd,OAAO,MAAM,KAAK,IAAI,OAAO;AAAA,QAC7B,eAAe,QAAQ,iBAAiB;AAAA,QACxC;AAAA,QACA,QAAQ,MAAM,MAAM,IAAI,OAAO;AAAA,QAC/B,aAAa,MAAM,WAAW,IAAI,OAAO;AAAA,QACzC;AAAA,QACA,KAAK,EAAE,OAAO;AAAA,MAChB;AAAA,IACF,QAAQ;AAAA,IAA4B;AAAA,EACtC;AACA,SAAO;AACT;AAEO,SAAS,cAAc,MAAc,SAAuF;AACjI,MAAI;AACF,UAAM,EAAE,SAAS,IAAI,UAAU,IAAI;AACnC,UAAM,OAAO,IAAI,IAAI,OAAO;AAC5B,WAAO,MAAM,KAAM,SAAiB,iBAAiB,KAAK,CAAC,EACxD,IAAI,CAAC,QAAa;AACjB,YAAM,MAAM,IAAI,aAAa,KAAK,KAAK,IAAI,aAAa,UAAU,KAAK;AACvE,UAAI,CAAC,IAAK,QAAO;AACjB,UAAI;AACF,eAAO;AAAA,UACL,KAAK,IAAI,IAAI,KAAK,IAAI,EAAE;AAAA,UACxB,KAAM,IAAI,aAAa,KAAK,KAAK;AAAA,UACjC,OAAO,IAAI,aAAa,OAAO,IAAI,SAAS,IAAI,aAAa,OAAO,CAAC,IAAI;AAAA,UACzE,QAAQ,IAAI,aAAa,QAAQ,IAAI,SAAS,IAAI,aAAa,QAAQ,CAAC,IAAI;AAAA,QAC9E;AAAA,MACF,QAAQ;AAAE,eAAO;AAAA,MAAM;AAAA,IACzB,CAAC,EACA,OAAO,CAAC,MAA0E,MAAM,IAAI;AAAA,EACjG,QAAQ;AACN,WAAO,CAAC;AAAA,EACV;AACF;AAEO,SAAS,kBAAkB,MAAc,SAAsG;AACpJ,QAAM,YAAY;AAClB,QAAM,YAAY;AAClB,QAAM,YAAY;AAClB,QAAM,UAAU;AAEhB,WAAS,SAAS,KAAa,KAAuD;AACpF,QAAI,QAAQ,SAAS,UAAU,KAAK,GAAG,EAAG,QAAO;AACjD,QAAI,QAAQ,WAAW,UAAU,KAAK,GAAG,EAAG,QAAO;AACnD,QAAI,QAAQ,WAAW,UAAU,KAAK,GAAG,EAAG,QAAO;AACnD,QAAI,QAAQ,KAAK,GAAG,EAAG,QAAO;AAC9B,WAAO;AAAA,EACT;AAEA,MAAI;AACF,UAAM,EAAE,SAAS,IAAI,UAAU,IAAI;AACnC,UAAM,OAAO,IAAI,IAAI,OAAO;AAC5B,UAAM,UAA+F,CAAC;AACtG,UAAM,OAAO,oBAAI,IAAY;AAE7B,UAAM,YAA8B,CAAC,8CAA8C,KAAK;AACxF,UAAM,YAAqC;AAAA,MACzC,CAAC,OAAO,KAAK;AAAA,MAAG,CAAC,SAAS,KAAK;AAAA,MAAG,CAAC,SAAS,KAAK;AAAA,MAAG,CAAC,UAAU,KAAK;AAAA,MACpE,CAAC,KAAK,MAAM;AAAA,IACd;AAEA,eAAW,CAAC,KAAK,IAAI,KAAK,WAAW;AACnC,iBAAW,MAAM,CAAC,GAAI,SAAiB,iBAAiB,GAAG,CAAC,GAAG;AAC7D,cAAM,MAAO,GAAW,aAAa,IAAI,KAAK;AAC9C,YAAI,CAAC,IAAK;AACV,YAAI;AACF,gBAAM,WAAW,IAAI,IAAI,KAAK,IAAI,EAAE;AACpC,cAAI,KAAK,IAAI,QAAQ,EAAG;AACxB,cAAI,QAAQ,OAAO,CAAC,QAAQ,KAAK,QAAQ,KAAK,CAAC,UAAU,KAAK,QAAQ,KAAK,CAAC,UAAU,KAAK,QAAQ,KAAK,CAAC,UAAU,KAAK,QAAQ,EAAG;AACnI,eAAK,IAAI,QAAQ;AACjB,kBAAQ,KAAK,EAAE,KAAK,UAAU,MAAM,SAAS,UAAU,GAAG,GAAG,KAAK,IAAI,CAAC;AAAA,QACzE,QAAQ;AAAA,QAAa;AAAA,MACvB;AAAA,IACF;AACA,WAAO;AAAA,EACT,QAAQ;AACN,WAAO,CAAC;AAAA,EACV;AACF;",
+  "sourcesContent": ["import { parseHTML } from 'linkedom';\n\nexport interface ExtractionResult {\n  matches: string[];\n  count: number;\n  method: string;\n}\n\nexport type TableRow = Record<string, string>;\n\nexport interface PageMetadata {\n  title: string | null;\n  description: string | null;\n  keywords: string | null;\n  ogTitle: string | null;\n  ogDescription: string | null;\n  ogImage: string | null;\n  ogUrl: string | null;\n  ogType: string | null;\n  ogSiteName: string | null;\n  ogLocale: string | null;\n  twitterCard: string | null;\n  twitterTitle: string | null;\n  twitterDescription: string | null;\n  twitterImage: string | null;\n  twitterSite: string | null;\n  canonical: string | null;\n  robots: string | null;\n  hreflang: Array<{ lang: string; href: string }>;\n  themeColor: string | null;\n  favicon: string | null;\n  jsonLd: object[];\n  schemaTypes: string[];\n  lang: string | null;\n}\n\nexport function extractByCssSelector(html: string, selector: string, attr?: string): ExtractionResult {\n  try {\n    const { document } = parseHTML(html);\n    const nodes = Array.from(document.querySelectorAll(selector));\n    const matches = nodes.map((n: any) => {\n      if (attr) return (n as any).getAttribute(attr) ?? '';\n      return (n as any).textContent?.trim() ?? '';\n    }).filter(Boolean);\n    return { matches, count: matches.length, method: 'css-selector' };\n  } catch (e) {\n    return { matches: [], count: 0, method: 'css-selector-error' };\n  }\n}\n\nexport function extractByXPath(html: string, xpath: string): ExtractionResult {\n  try {\n    const { document, XPathResult } = parseHTML(html) as any;\n    const result = document.evaluate(xpath, document, null, XPathResult?.ANY_TYPE ?? 0, null);\n    const matches: string[] = [];\n    let node = result.iterateNext?.();\n    while (node) {\n      const text = (node as any).textContent?.trim() ?? (node as any).nodeValue?.trim() ?? '';\n      if (text) matches.push(text);\n      node = result.iterateNext?.();\n    }\n    return { matches, count: matches.length, method: 'xpath' };\n  } catch (e) {\n    return { matches: [], count: 0, method: 'xpath-error' };\n  }\n}\n\nexport function extractByRegex(text: string, pattern: string, flags?: string): ExtractionResult {\n  try {\n    const re = new RegExp(pattern, (flags ?? 'g').includes('g') ? flags : (flags ?? '') + 'g');\n    const matches: string[] = [];\n    let m: RegExpExecArray | null;\n    while ((m = re.exec(text)) !== null) {\n      matches.push(m[1] ?? m[0]);\n      if (!re.global) break;\n    }\n    return { matches, count: matches.length, method: 'regex' };\n  } catch {\n    return { matches: [], count: 0, method: 'regex-error' };\n  }\n}\n\nexport function extractTables(html: string): TableRow[][] {\n  try {\n    const { document } = parseHTML(html);\n    const tables = Array.from(document.querySelectorAll('table'));\n    return tables.map((table: any) => {\n      const headers: string[] = [];\n      const rows: TableRow[] = [];\n      const ths = [...table.querySelectorAll('thead th, thead td, tr:first-child th')];\n      if (ths.length > 0) {\n        ths.forEach((th: any, i: number) => headers.push(th.textContent?.trim() || `col${i + 1}`));\n      }\n      const trs = [...table.querySelectorAll('tbody tr, tr')];\n      for (const tr of trs) {\n        const cells = [...(tr as any).querySelectorAll('td, th')];\n        if (cells.length === 0) continue;\n        if (headers.length === 0) {\n          cells.forEach((_: any, i: number) => headers.push(`col${i + 1}`));\n        }\n        const row: TableRow = {};\n        cells.forEach((cell: any, i: number) => {\n          const key = headers[i] || `col${i + 1}`;\n          row[key] = cell.textContent?.trim() ?? '';\n        });\n        if (Object.values(row).some(v => v)) rows.push(row);\n      }\n      return rows;\n    }).filter(t => t.length > 0);\n  } catch {\n    return [];\n  }\n}\n\nexport function extractMetadata(html: string, url: string): PageMetadata {\n  const meta: PageMetadata = {\n    title: null, description: null, keywords: null,\n    ogTitle: null, ogDescription: null, ogImage: null, ogUrl: null, ogType: null,\n    ogSiteName: null, ogLocale: null,\n    twitterCard: null, twitterTitle: null, twitterDescription: null,\n    twitterImage: null, twitterSite: null,\n    canonical: null, robots: null,\n    hreflang: [], themeColor: null, favicon: null,\n    jsonLd: [], schemaTypes: [], lang: null,\n  };\n  try {\n    const { document } = parseHTML(html);\n    meta.title = (document as any).title ?? (document as any).querySelector('title')?.textContent?.trim() ?? null;\n    meta.lang = (document as any).documentElement?.getAttribute('lang') ?? null;\n\n    const metas = [...(document as any).querySelectorAll('meta')];\n    for (const m of metas) {\n      const name = (m as any).getAttribute('name')?.toLowerCase() ?? '';\n      const prop = (m as any).getAttribute('property')?.toLowerCase() ?? '';\n      const content = (m as any).getAttribute('content') ?? '';\n      if (name === 'description') meta.description = content;\n      if (name === 'keywords') meta.keywords = content;\n      if (name === 'robots') meta.robots = content;\n      if (name === 'twitter:card') meta.twitterCard = content;\n      if (name === 'twitter:title') meta.twitterTitle = content;\n      if (name === 'twitter:description') meta.twitterDescription = content;\n      if (name === 'twitter:image') meta.twitterImage = content;\n      if (name === 'twitter:site') meta.twitterSite = content;\n      if (name === 'theme-color') meta.themeColor = content;\n      if (prop === 'og:title') meta.ogTitle = content;\n      if (prop === 'og:description') meta.ogDescription = content;\n      if (prop === 'og:image') meta.ogImage = content;\n      if (prop === 'og:url') meta.ogUrl = content;\n      if (prop === 'og:type') meta.ogType = content;\n      if (prop === 'og:site_name') meta.ogSiteName = content;\n      if (prop === 'og:locale') meta.ogLocale = content;\n    }\n\n    const canonical = (document as any).querySelector('link[rel=\"canonical\"]');\n    if (canonical) meta.canonical = canonical.getAttribute('href') ?? null;\n\n    const faviconEl = (document as any).querySelector('link[rel=\"icon\"], link[rel=\"shortcut icon\"]');\n    if (faviconEl) meta.favicon = faviconEl.getAttribute('href') ?? null;\n\n    const hreflangEls = [...(document as any).querySelectorAll('link[rel=\"alternate\"][hreflang]')];\n    for (const el of hreflangEls) {\n      const lang = (el as any).getAttribute('hreflang') ?? '';\n      const href = (el as any).getAttribute('href') ?? '';\n      if (lang && href) meta.hreflang.push({ lang, href });\n    }\n\n    const jsonLdScripts = [...(document as any).querySelectorAll('script[type=\"application/ld+json\"]')];\n    for (const s of jsonLdScripts) {\n      try {\n        const parsed = JSON.parse((s as any).textContent ?? '{}');\n        meta.jsonLd.push(parsed);\n        const types = Array.isArray(parsed) ? parsed.map((p: any) => p['@type']).flat() : [parsed['@type']];\n        meta.schemaTypes.push(...types.filter(Boolean));\n      } catch { /* skip invalid JSON-LD */ }\n    }\n  } catch { /* non-fatal */ }\n  return meta;\n}\n\nexport interface ProductData {\n  name: string | null;\n  description: string | null;\n  brand: string | null;\n  sku: string | null;\n  price: number | null;\n  priceCurrency: string | null;\n  availability: string | null;\n  rating: number | null;\n  reviewCount: number | null;\n  image: string | null;\n  url: string | null;\n}\n\nexport function extractStructuredProduct(jsonLd: object[]): ProductData | null {\n  for (const item of jsonLd) {\n    try {\n      const type = (item as any)['@type'];\n      const isProduct = Array.isArray(type)\n        ? type.some((t: any) => typeof t === 'string' && t.toLowerCase() === 'product')\n        : typeof type === 'string' && type.toLowerCase() === 'product';\n      if (!isProduct) continue;\n\n      const i = item as any;\n      const offers = Array.isArray(i.offers) ? i.offers[0] : i.offers;\n\n      const rawPrice = offers?.price;\n      const price = rawPrice != null ? parseFloat(rawPrice) : NaN;\n\n      const rawRating = i.aggregateRating?.ratingValue;\n      const rating = rawRating != null ? parseFloat(rawRating) : NaN;\n\n      const rawReviewCount = i.aggregateRating?.reviewCount;\n      const reviewCount = rawReviewCount != null ? parseInt(rawReviewCount, 10) : NaN;\n\n      const rawAvailability: string | undefined = offers?.availability;\n      const availability = rawAvailability\n        ? rawAvailability.replace(/^https?:\\/\\/schema\\.org\\//, '')\n        : null;\n\n      const rawImage = i.image;\n      const image = Array.isArray(rawImage) ? (rawImage[0] ?? null) : (rawImage ?? null);\n\n      const brand = i.brand?.name ?? i.brand ?? null;\n\n      return {\n        name: i.name ?? null,\n        description: i.description ?? null,\n        brand: typeof brand === 'string' ? brand : null,\n        sku: i.sku ?? null,\n        price: isNaN(price) ? null : price,\n        priceCurrency: offers?.priceCurrency ?? null,\n        availability,\n        rating: isNaN(rating) ? null : rating,\n        reviewCount: isNaN(reviewCount) ? null : reviewCount,\n        image,\n        url: i.url ?? null,\n      };\n    } catch { /* skip malformed item */ }\n  }\n  return null;\n}\n\nexport function extractImages(html: string, baseUrl: string): Array<{ src: string; alt: string; width?: number; height?: number }> {\n  try {\n    const { document } = parseHTML(html);\n    const base = new URL(baseUrl);\n    return Array.from((document as any).querySelectorAll('img'))\n      .map((img: any) => {\n        const src = img.getAttribute('src') ?? img.getAttribute('data-src') ?? '';\n        if (!src) return null;\n        try {\n          return {\n            src: new URL(src, base).href,\n            alt: (img.getAttribute('alt') ?? '') as string,\n            width: img.getAttribute('width') ? parseInt(img.getAttribute('width')) : undefined,\n            height: img.getAttribute('height') ? parseInt(img.getAttribute('height')) : undefined,\n          } as { src: string; alt: string; width?: number; height?: number };\n        } catch { return null; }\n      })\n      .filter((x): x is { src: string; alt: string; width?: number; height?: number } => x !== null);\n  } catch {\n    return [];\n  }\n}\n\nexport function extractMediaFiles(html: string, baseUrl: string): Array<{ url: string; type: 'image' | 'video' | 'audio' | 'document'; tag: string }> {\n  const IMAGE_EXT = /\\.(png|jpe?g|gif|webp|svg|ico|avif)(\\?.*)?$/i;\n  const VIDEO_EXT = /\\.(mp4|webm|ogv|mov|avi)(\\?.*)?$/i;\n  const AUDIO_EXT = /\\.(mp3|ogg|wav|flac|aac)(\\?.*)?$/i;\n  const DOC_EXT = /\\.(pdf|docx?|xlsx?|pptx?|csv|zip)(\\?.*)?$/i;\n\n  function classify(url: string, tag: string): 'image' | 'video' | 'audio' | 'document' {\n    if (tag === 'img' || IMAGE_EXT.test(url)) return 'image';\n    if (tag === 'video' || VIDEO_EXT.test(url)) return 'video';\n    if (tag === 'audio' || AUDIO_EXT.test(url)) return 'audio';\n    if (DOC_EXT.test(url)) return 'document';\n    return 'document';\n  }\n\n  try {\n    const { document } = parseHTML(html);\n    const base = new URL(baseUrl);\n    const results: Array<{ url: string; type: 'image' | 'video' | 'audio' | 'document'; tag: string }> = [];\n    const seen = new Set<string>();\n\n    const selectors: [string, string] = ['img[src],video[src],audio[src],source[src]', 'src'];\n    const attrPairs: Array<[string, string]> = [\n      ['img', 'src'], ['video', 'src'], ['audio', 'src'], ['source', 'src'],\n      ['a', 'href'],\n    ];\n\n    for (const [sel, attr] of attrPairs) {\n      for (const el of [...(document as any).querySelectorAll(sel)]) {\n        const raw = (el as any).getAttribute(attr) ?? '';\n        if (!raw) continue;\n        try {\n          const resolved = new URL(raw, base).href;\n          if (seen.has(resolved)) continue;\n          if (sel === 'a' && !DOC_EXT.test(resolved) && !IMAGE_EXT.test(resolved) && !VIDEO_EXT.test(resolved) && !AUDIO_EXT.test(resolved)) continue;\n          seen.add(resolved);\n          results.push({ url: resolved, type: classify(resolved, sel), tag: sel });\n        } catch { /* skip */ }\n      }\n    }\n    return results;\n  } catch {\n    return [];\n  }\n}\n"],
+  "mappings": "AAAA,SAAS,iBAAiB;AAoCnB,SAAS,qBAAqB,MAAc,UAAkB,MAAiC;AACpG,MAAI;AACF,UAAM,EAAE,SAAS,IAAI,UAAU,IAAI;AACnC,UAAM,QAAQ,MAAM,KAAK,SAAS,iBAAiB,QAAQ,CAAC;AAC5D,UAAM,UAAU,MAAM,IAAI,CAAC,MAAW;AACpC,UAAI,KAAM,QAAQ,EAAU,aAAa,IAAI,KAAK;AAClD,aAAQ,EAAU,aAAa,KAAK,KAAK;AAAA,IAC3C,CAAC,EAAE,OAAO,OAAO;AACjB,WAAO,EAAE,SAAS,OAAO,QAAQ,QAAQ,QAAQ,eAAe;AAAA,EAClE,SAAS,GAAG;AACV,WAAO,EAAE,SAAS,CAAC,GAAG,OAAO,GAAG,QAAQ,qBAAqB;AAAA,EAC/D;AACF;AAEO,SAAS,eAAe,MAAc,OAAiC;AAC5E,MAAI;AACF,UAAM,EAAE,UAAU,YAAY,IAAI,UAAU,IAAI;AAChD,UAAM,SAAS,SAAS,SAAS,OAAO,UAAU,MAAM,aAAa,YAAY,GAAG,IAAI;AACxF,UAAM,UAAoB,CAAC;AAC3B,QAAI,OAAO,OAAO,cAAc;AAChC,WAAO,MAAM;AACX,YAAM,OAAQ,KAAa,aAAa,KAAK,KAAM,KAAa,WAAW,KAAK,KAAK;AACrF,UAAI,KAAM,SAAQ,KAAK,IAAI;AAC3B,aAAO,OAAO,cAAc;AAAA,IAC9B;AACA,WAAO,EAAE,SAAS,OAAO,QAAQ,QAAQ,QAAQ,QAAQ;AAAA,EAC3D,SAAS,GAAG;AACV,WAAO,EAAE,SAAS,CAAC,GAAG,OAAO,GAAG,QAAQ,cAAc;AAAA,EACxD;AACF;AAEO,SAAS,eAAe,MAAc,SAAiB,OAAkC;AAC9F,MAAI;AACF,UAAM,KAAK,IAAI,OAAO,UAAU,SAAS,KAAK,SAAS,GAAG,IAAI,SAAS,SAAS,MAAM,GAAG;AACzF,UAAM,UAAoB,CAAC;AAC3B,QAAI;AACJ,YAAQ,IAAI,GAAG,KAAK,IAAI,OAAO,MAAM;AACnC,cAAQ,KAAK,EAAE,CAAC,KAAK,EAAE,CAAC,CAAC;AACzB,UAAI,CAAC,GAAG,OAAQ;AAAA,IAClB;AACA,WAAO,EAAE,SAAS,OAAO,QAAQ,QAAQ,QAAQ,QAAQ;AAAA,EAC3D,QAAQ;AACN,WAAO,EAAE,SAAS,CAAC,GAAG,OAAO,GAAG,QAAQ,cAAc;AAAA,EACxD;AACF;AAEO,SAAS,cAAc,MAA4B;AACxD,MAAI;AACF,UAAM,EAAE,SAAS,IAAI,UAAU,IAAI;AACnC,UAAM,SAAS,MAAM,KAAK,SAAS,iBAAiB,OAAO,CAAC;AAC5D,WAAO,OAAO,IAAI,CAAC,UAAe;AAChC,YAAM,UAAoB,CAAC;AAC3B,YAAM,OAAmB,CAAC;AAC1B,YAAM,MAAM,CAAC,GAAG,MAAM,iBAAiB,uCAAuC,CAAC;AAC/E,UAAI,IAAI,SAAS,GAAG;AAClB,YAAI,QAAQ,CAAC,IAAS,MAAc,QAAQ,KAAK,GAAG,aAAa,KAAK,KAAK,MAAM,IAAI,CAAC,EAAE,CAAC;AAAA,MAC3F;AACA,YAAM,MAAM,CAAC,GAAG,MAAM,iBAAiB,cAAc,CAAC;AACtD,iBAAW,MAAM,KAAK;AACpB,cAAM,QAAQ,CAAC,GAAI,GAAW,iBAAiB,QAAQ,CAAC;AACxD,YAAI,MAAM,WAAW,EAAG;AACxB,YAAI,QAAQ,WAAW,GAAG;AACxB,gBAAM,QAAQ,CAAC,GAAQ,MAAc,QAAQ,KAAK,MAAM,IAAI,CAAC,EAAE,CAAC;AAAA,QAClE;AACA,cAAM,MAAgB,CAAC;AACvB,cAAM,QAAQ,CAAC,MAAW,MAAc;AACtC,gBAAM,MAAM,QAAQ,CAAC,KAAK,MAAM,IAAI,CAAC;AACrC,cAAI,GAAG,IAAI,KAAK,aAAa,KAAK,KAAK;AAAA,QACzC,CAAC;AACD,YAAI,OAAO,OAAO,GAAG,EAAE,KAAK,OAAK,CAAC,EAAG,MAAK,KAAK,GAAG;AAAA,MACpD;AACA,aAAO;AAAA,IACT,CAAC,EAAE,OAAO,OAAK,EAAE,SAAS,CAAC;AAAA,EAC7B,QAAQ;AACN,WAAO,CAAC;AAAA,EACV;AACF;AAEO,SAAS,gBAAgB,MAAc,KAA2B;AACvE,QAAM,OAAqB;AAAA,IACzB,OAAO;AAAA,IAAM,aAAa;AAAA,IAAM,UAAU;AAAA,IAC1C,SAAS;AAAA,IAAM,eAAe;AAAA,IAAM,SAAS;AAAA,IAAM,OAAO;AAAA,IAAM,QAAQ;AAAA,IACxE,YAAY;AAAA,IAAM,UAAU;AAAA,IAC5B,aAAa;AAAA,IAAM,cAAc;AAAA,IAAM,oBAAoB;AAAA,IAC3D,cAAc;AAAA,IAAM,aAAa;AAAA,IACjC,WAAW;AAAA,IAAM,QAAQ;AAAA,IACzB,UAAU,CAAC;AAAA,IAAG,YAAY;AAAA,IAAM,SAAS;AAAA,IACzC,QAAQ,CAAC;AAAA,IAAG,aAAa,CAAC;AAAA,IAAG,MAAM;AAAA,EACrC;AACA,MAAI;AACF,UAAM,EAAE,SAAS,IAAI,UAAU,IAAI;AACnC,SAAK,QAAS,SAAiB,SAAU,SAAiB,cAAc,OAAO,GAAG,aAAa,KAAK,KAAK;AACzG,SAAK,OAAQ,SAAiB,iBAAiB,aAAa,MAAM,KAAK;AAEvE,UAAM,QAAQ,CAAC,GAAI,SAAiB,iBAAiB,MAAM,CAAC;AAC5D,eAAW,KAAK,OAAO;AACrB,YAAM,OAAQ,EAAU,aAAa,MAAM,GAAG,YAAY,KAAK;AAC/D,YAAM,OAAQ,EAAU,aAAa,UAAU,GAAG,YAAY,KAAK;AACnE,YAAM,UAAW,EAAU,aAAa,SAAS,KAAK;AACtD,UAAI,SAAS,cAAe,MAAK,cAAc;AAC/C,UAAI,SAAS,WAAY,MAAK,WAAW;AACzC,UAAI,SAAS,SAAU,MAAK,SAAS;AACrC,UAAI,SAAS,eAAgB,MAAK,cAAc;AAChD,UAAI,SAAS,gBAAiB,MAAK,eAAe;AAClD,UAAI,SAAS,sBAAuB,MAAK,qBAAqB;AAC9D,UAAI,SAAS,gBAAiB,MAAK,eAAe;AAClD,UAAI,SAAS,eAAgB,MAAK,cAAc;AAChD,UAAI,SAAS,cAAe,MAAK,aAAa;AAC9C,UAAI,SAAS,WAAY,MAAK,UAAU;AACxC,UAAI,SAAS,iBAAkB,MAAK,gBAAgB;AACpD,UAAI,SAAS,WAAY,MAAK,UAAU;AACxC,UAAI,SAAS,SAAU,MAAK,QAAQ;AACpC,UAAI,SAAS,UAAW,MAAK,SAAS;AACtC,UAAI,SAAS,eAAgB,MAAK,aAAa;AAC/C,UAAI,SAAS,YAAa,MAAK,WAAW;AAAA,IAC5C;AAEA,UAAM,YAAa,SAAiB,cAAc,uBAAuB;AACzE,QAAI,UAAW,MAAK,YAAY,UAAU,aAAa,MAAM,KAAK;AAElE,UAAM,YAAa,SAAiB,cAAc,6CAA6C;AAC/F,QAAI,UAAW,MAAK,UAAU,UAAU,aAAa,MAAM,KAAK;AAEhE,UAAM,cAAc,CAAC,GAAI,SAAiB,iBAAiB,iCAAiC,CAAC;AAC7F,eAAW,MAAM,aAAa;AAC5B,YAAM,OAAQ,GAAW,aAAa,UAAU,KAAK;AACrD,YAAM,OAAQ,GAAW,aAAa,MAAM,KAAK;AACjD,UAAI,QAAQ,KAAM,MAAK,SAAS,KAAK,EAAE,MAAM,KAAK,CAAC;AAAA,IACrD;AAEA,UAAM,gBAAgB,CAAC,GAAI,SAAiB,iBAAiB,oCAAoC,CAAC;AAClG,eAAW,KAAK,eAAe;AAC7B,UAAI;AACF,cAAM,SAAS,KAAK,MAAO,EAAU,eAAe,IAAI;AACxD,aAAK,OAAO,KAAK,MAAM;AACvB,cAAM,QAAQ,MAAM,QAAQ,MAAM,IAAI,OAAO,IAAI,CAAC,MAAW,EAAE,OAAO,CAAC,EAAE,KAAK,IAAI,CAAC,OAAO,OAAO,CAAC;AAClG,aAAK,YAAY,KAAK,GAAG,MAAM,OAAO,OAAO,CAAC;AAAA,MAChD,QAAQ;AAAA,MAA6B;AAAA,IACvC;AAAA,EACF,QAAQ;AAAA,EAAkB;AAC1B,SAAO;AACT;AAgBO,SAAS,yBAAyB,QAAsC;AAC7E,aAAW,QAAQ,QAAQ;AACzB,QAAI;AACF,YAAM,OAAQ,KAAa,OAAO;AAClC,YAAM,YAAY,MAAM,QAAQ,IAAI,IAChC,KAAK,KAAK,CAAC,MAAW,OAAO,MAAM,YAAY,EAAE,YAAY,MAAM,SAAS,IAC5E,OAAO,SAAS,YAAY,KAAK,YAAY,MAAM;AACvD,UAAI,CAAC,UAAW;AAEhB,YAAM,IAAI;AACV,YAAM,SAAS,MAAM,QAAQ,EAAE,MAAM,IAAI,EAAE,OAAO,CAAC,IAAI,EAAE;AAEzD,YAAM,WAAW,QAAQ;AACzB,YAAM,QAAQ,YAAY,OAAO,WAAW,QAAQ,IAAI;AAExD,YAAM,YAAY,EAAE,iBAAiB;AACrC,YAAM,SAAS,aAAa,OAAO,WAAW,SAAS,IAAI;AAE3D,YAAM,iBAAiB,EAAE,iBAAiB;AAC1C,YAAM,cAAc,kBAAkB,OAAO,SAAS,gBAAgB,EAAE,IAAI;AAE5E,YAAM,kBAAsC,QAAQ;AACpD,YAAM,eAAe,kBACjB,gBAAgB,QAAQ,6BAA6B,EAAE,IACvD;AAEJ,YAAM,WAAW,EAAE;AACnB,YAAM,QAAQ,MAAM,QAAQ,QAAQ,IAAK,SAAS,CAAC,KAAK,OAAS,YAAY;AAE7E,YAAM,QAAQ,EAAE,OAAO,QAAQ,EAAE,SAAS;AAE1C,aAAO;AAAA,QACL,MAAM,EAAE,QAAQ;AAAA,QAChB,aAAa,EAAE,eAAe;AAAA,QAC9B,OAAO,OAAO,UAAU,WAAW,QAAQ;AAAA,QAC3C,KAAK,EAAE,OAAO;AAAA,QACd,OAAO,MAAM,KAAK,IAAI,OAAO;AAAA,QAC7B,eAAe,QAAQ,iBAAiB;AAAA,QACxC;AAAA,QACA,QAAQ,MAAM,MAAM,IAAI,OAAO;AAAA,QAC/B,aAAa,MAAM,WAAW,IAAI,OAAO;AAAA,QACzC;AAAA,QACA,KAAK,EAAE,OAAO;AAAA,MAChB;AAAA,IACF,QAAQ;AAAA,IAA4B;AAAA,EACtC;AACA,SAAO;AACT;AAEO,SAAS,cAAc,MAAc,SAAuF;AACjI,MAAI;AACF,UAAM,EAAE,SAAS,IAAI,UAAU,IAAI;AACnC,UAAM,OAAO,IAAI,IAAI,OAAO;AAC5B,WAAO,MAAM,KAAM,SAAiB,iBAAiB,KAAK,CAAC,EACxD,IAAI,CAAC,QAAa;AACjB,YAAM,MAAM,IAAI,aAAa,KAAK,KAAK,IAAI,aAAa,UAAU,KAAK;AACvE,UAAI,CAAC,IAAK,QAAO;AACjB,UAAI;AACF,eAAO;AAAA,UACL,KAAK,IAAI,IAAI,KAAK,IAAI,EAAE;AAAA,UACxB,KAAM,IAAI,aAAa,KAAK,KAAK;AAAA,UACjC,OAAO,IAAI,aAAa,OAAO,IAAI,SAAS,IAAI,aAAa,OAAO,CAAC,IAAI;AAAA,UACzE,QAAQ,IAAI,aAAa,QAAQ,IAAI,SAAS,IAAI,aAAa,QAAQ,CAAC,IAAI;AAAA,QAC9E;AAAA,MACF,QAAQ;AAAE,eAAO;AAAA,MAAM;AAAA,IACzB,CAAC,EACA,OAAO,CAAC,MAA0E,MAAM,IAAI;AAAA,EACjG,QAAQ;AACN,WAAO,CAAC;AAAA,EACV;AACF;AAEO,SAAS,kBAAkB,MAAc,SAAsG;AACpJ,QAAM,YAAY;AAClB,QAAM,YAAY;AAClB,QAAM,YAAY;AAClB,QAAM,UAAU;AAEhB,WAAS,SAAS,KAAa,KAAuD;AACpF,QAAI,QAAQ,SAAS,UAAU,KAAK,GAAG,EAAG,QAAO;AACjD,QAAI,QAAQ,WAAW,UAAU,KAAK,GAAG,EAAG,QAAO;AACnD,QAAI,QAAQ,WAAW,UAAU,KAAK,GAAG,EAAG,QAAO;AACnD,QAAI,QAAQ,KAAK,GAAG,EAAG,QAAO;AAC9B,WAAO;AAAA,EACT;AAEA,MAAI;AACF,UAAM,EAAE,SAAS,IAAI,UAAU,IAAI;AACnC,UAAM,OAAO,IAAI,IAAI,OAAO;AAC5B,UAAM,UAA+F,CAAC;AACtG,UAAM,OAAO,oBAAI,IAAY;AAE7B,UAAM,YAA8B,CAAC,8CAA8C,KAAK;AACxF,UAAM,YAAqC;AAAA,MACzC,CAAC,OAAO,KAAK;AAAA,MAAG,CAAC,SAAS,KAAK;AAAA,MAAG,CAAC,SAAS,KAAK;AAAA,MAAG,CAAC,UAAU,KAAK;AAAA,MACpE,CAAC,KAAK,MAAM;AAAA,IACd;AAEA,eAAW,CAAC,KAAK,IAAI,KAAK,WAAW;AACnC,iBAAW,MAAM,CAAC,GAAI,SAAiB,iBAAiB,GAAG,CAAC,GAAG;AAC7D,cAAM,MAAO,GAAW,aAAa,IAAI,KAAK;AAC9C,YAAI,CAAC,IAAK;AACV,YAAI;AACF,gBAAM,WAAW,IAAI,IAAI,KAAK,IAAI,EAAE;AACpC,cAAI,KAAK,IAAI,QAAQ,EAAG;AACxB,cAAI,QAAQ,OAAO,CAAC,QAAQ,KAAK,QAAQ,KAAK,CAAC,UAAU,KAAK,QAAQ,KAAK,CAAC,UAAU,KAAK,QAAQ,KAAK,CAAC,UAAU,KAAK,QAAQ,EAAG;AACnI,eAAK,IAAI,QAAQ;AACjB,kBAAQ,KAAK,EAAE,KAAK,UAAU,MAAM,SAAS,UAAU,GAAG,GAAG,KAAK,IAAI,CAAC;AAAA,QACzE,QAAQ;AAAA,QAAa;AAAA,MACvB;AAAA,IACF;AACA,WAAO;AAAA,EACT,QAAQ;AACN,WAAO,CAAC;AAAA,EACV;AACF;",
   "names": []
 }

package/dist/interactive-probe.js

@@ -106,7 +106,7 @@ async function findModalTriggers(page) {
         p = p.parentElement;
       }
       return true;
-    }).slice(0, 8).map((el) => {
+    }).slice(0, 20).map((el) => {
       const label = el.innerText?.trim() || el.getAttribute("aria-label") || "button";
       return { selector: `[aria-label="${label}"]`, label: label.slice(0, 60) };
     });
@@ -161,7 +161,20 @@ async function probeScreen(page, screenId, url, elements) {
         await submitBtn.click({ timeout: 2e3 });
         await page.waitForTimeout(1500);
         const errors = await collectErrors(page);
-        for (const errorText of errors) {
+        const ariaFieldErrors = await page.evaluate(
+          () => Array.from(document.querySelectorAll('[aria-invalid="true"]')).map((el) => {
+            const describedBy = el.getAttribute("aria-describedby");
+            const errorEl = describedBy ? document.getElementById(describedBy) : null;
+            const label = el.getAttribute("aria-label") || el.getAttribute("placeholder") || el.name || "";
+            return { label, errorText: errorEl?.textContent?.trim() || "" };
+          }).filter((f) => f.errorText)
+        );
+        const ariaMatched = /* @__PURE__ */ new Set();
+        for (const af of ariaFieldErrors) {
+          ariaMatched.add(af.errorText);
+          result.validationErrors.push({ field: af.label || "unknown", trigger: "empty_submit", errorText: af.errorText.slice(0, 150) });
+        }
+        for (const errorText of errors.filter((e) => !ariaMatched.has(e))) {
           const matchedField = formInputs.find(
             (el) => errorText.toLowerCase().includes(el.meaning.toLowerCase().split(" ")[0]) || el.meaning.toLowerCase().includes(errorText.toLowerCase().split(" ")[0])
           );
@@ -240,6 +253,28 @@ async function probeScreen(page, screenId, url, elements) {
             content: dialogText.content,
             fields: dialogText.fields.slice(0, 10)
           });
+        } else {
+          const accordionText = await page.evaluate(() => {
+            const panels = Array.from(document.querySelectorAll(
+              '[aria-expanded="true"] + *, details[open] > :not(summary), [data-state="open"]:not([role="dialog"]):not([class*="modal" i]), [data-headlessui-state="open"] > *'
+            )).filter((el) => {
+              const s = window.getComputedStyle(el);
+              return s.display !== "none" && s.visibility !== "hidden";
+            });
+            if (panels.length === 0) return null;
+            const panel = panels[0];
+            const heading = panel.querySelector("h2,h3,h4")?.textContent?.replace(/\s+/g, " ").trim() || "";
+            const content = panel.innerText?.replace(/\s+/g, " ").trim().slice(0, 400) || "";
+            return content.length > 15 ? { heading, content } : null;
+          });
+          if (accordionText) {
+            result.modals.push({
+              triggerElement: trigger.label,
+              title: accordionText.heading || `${trigger.label} (revealed)`,
+              content: accordionText.content,
+              fields: []
+            });
+          }
         }
         await page.keyboard.press("Escape");
         await page.waitForTimeout(400);
@@ -257,6 +292,34 @@ async function probeScreen(page, screenId, url, elements) {
     result.conditionals.push(...conditionals);
   } catch {
   }
+  try {
+    result.tabPanels = await probeTabPanels(page);
+  } catch {
+  }
+  try {
+    result.tooltips = await probeTooltips(page);
+  } catch {
+  }
+  try {
+    result.carouselSlides = await probeCarousels(page);
+  } catch {
+  }
+  try {
+    result.wizardSteps = await probeWizardSteps(page);
+  } catch {
+  }
+  try {
+    result.focusTraps = await detectFocusTraps(page);
+  } catch {
+  }
+  try {
+    result.skipNavIssues = await probeSkipNavigation(page);
+  } catch {
+  }
+  try {
+    result.contextMenus = await probeContextMenus(page);
+  } catch {
+  }
   return result;
 }
 function probeResultToObservations(result) {
@@ -276,6 +339,27 @@ function probeResultToObservations(result) {
   for (const lp of result.loadingPatterns) {
     obs.push({ elementMeaning: "(submit button)", observationType: "LOADING_STATE", observedText: lp, triggeredBy: "form_submit" });
   }
+  for (const tp of result.tabPanels ?? []) {
+    obs.push({ elementMeaning: tp.tabLabel, observationType: "TAB_PANEL", observedText: tp.panelContent, triggeredBy: `tab_click:${tp.tabLabel}` });
+  }
+  for (const tt of result.tooltips ?? []) {
+    obs.push({ elementMeaning: tt.triggerLabel, observationType: "TOOLTIP", observedText: tt.tooltipText, triggeredBy: "hover_or_title" });
+  }
+  for (const cs of result.carouselSlides ?? []) {
+    obs.push({ elementMeaning: `Slide ${cs.slideIndex + 1}`, observationType: "CAROUSEL_SLIDE", observedText: cs.content, triggeredBy: `carousel_next:${cs.slideIndex}` });
+  }
+  for (const ws of result.wizardSteps ?? []) {
+    obs.push({ elementMeaning: ws.stepLabel ?? `Step ${ws.stepIndex + 1}`, observationType: "WIZARD_STEP", observedText: ws.content, triggeredBy: `wizard_next:${ws.stepIndex}` });
+  }
+  for (const ft of result.focusTraps ?? []) {
+    obs.push({ elementMeaning: ft.modalTrigger, observationType: "FOCUS_TRAP", observedText: ft.focusTrapped ? "Focus correctly trapped inside dialog" : "FAIL: focus escaped dialog \u2014 WCAG 2.1.2 violation", triggeredBy: "tab_key_sequence" });
+  }
+  for (const issue of result.skipNavIssues ?? []) {
+    obs.push({ elementMeaning: "(page)", observationType: "SKIP_NAV_ISSUE", observedText: issue, triggeredBy: "skip_nav_audit" });
+  }
+  for (const cm of result.contextMenus ?? []) {
+    obs.push({ elementMeaning: cm.trigger, observationType: "CONTEXT_MENU", observedText: cm.items.join(" | "), triggeredBy: "right_click" });
+  }
   return obs;
 }
 function probeResultToMarkdown(result) {
@@ -302,6 +386,39 @@ function probeResultToMarkdown(result) {
       if (m.fields.length > 0) lines.push(`  - Fields: ${m.fields.join(", ")}`);
     }
   }
+  if ((result.tabPanels ?? []).length > 0) {
+    lines.push("\n### Tab Panel Contents (observed)");
+    for (const tp of result.tabPanels) {
+      lines.push(`- **Tab: ${tp.tabLabel}**${tp.panelHeading ? ` \u2014 ${tp.panelHeading}` : ""}: ${tp.panelContent.slice(0, 200)}`);
+    }
+  }
+  if ((result.tooltips ?? []).length > 0) {
+    lines.push("\n### Tooltips (observed)");
+    for (const tt of result.tooltips) {
+      lines.push(`- **${tt.triggerLabel}**: ${tt.tooltipText}`);
+    }
+  }
+  if ((result.carouselSlides ?? []).length > 0) {
+    lines.push("\n### Carousel Slides (observed)");
+    for (const cs of result.carouselSlides) {
+      lines.push(`- **Slide ${cs.slideIndex + 1}**${cs.heading ? ` \u2014 ${cs.heading}` : ""}: ${cs.content.slice(0, 200)}`);
+    }
+  }
+  if ((result.wizardSteps ?? []).length > 0) {
+    lines.push("\n### Wizard / Multi-Step Form (observed)");
+    for (const ws of result.wizardSteps) {
+      const label = ws.stepLabel ? ` \u2014 ${ws.stepLabel}` : "";
+      lines.push(`- **Step ${ws.stepIndex + 1}${label}**: ${ws.content.slice(0, 200)}`);
+      if (ws.fields.length > 0) lines.push(`  - Fields: ${ws.fields.join(", ")}`);
+    }
+  }
+  if ((result.focusTraps ?? []).length > 0) {
+    lines.push("\n### Focus-Trap Audit (observed)");
+    for (const ft of result.focusTraps) {
+      const verdict = ft.focusTrapped ? "\u2705 PASS" : "\u274C FAIL (WCAG 2.1.2)";
+      lines.push(`- **${ft.modalTrigger}**: ${verdict} \u2014 focused: [${ft.sampledFocused.slice(0, 4).join(", ")}]`);
+    }
+  }
   if (result.conditionals.length > 0) {
     lines.push("\n### Conditional UI (observed)");
     for (const c of result.conditionals) {
@@ -383,6 +500,221 @@ ${interactionChunks.join("\n\n")}`;
   await onProgress?.(98, "Interactive probe complete");
   console.log(`[probe] Probed ${probeable.length} screens. Errors: ${errorCatalogueChunks.length}, Interactions: ${interactionChunks.length}`);
 }
+async function probeSkipNavigation(page) {
+  const issues = [];
+  try {
+    const skipLinks = await page.evaluate(
+      () => Array.from(document.querySelectorAll('a[href^="#"]')).filter((el) => /skip|jump to|main content|content|nav/i.test(el.innerText?.toLowerCase().trim() || "")).map((el) => ({
+        text: el.innerText?.trim() || "",
+        href: el.getAttribute("href") || "",
+        hasTarget: !!document.querySelector(el.getAttribute("href") || "#__invalid__")
+      }))
+    );
+    if (skipLinks.length === 0) {
+      issues.push("No skip navigation links found (WCAG 2.4.1 \u2014 keyboard users cannot bypass nav blocks)");
+    }
+    for (const link of skipLinks) {
+      if (!link.hasTarget) issues.push(`Skip link "${link.text}" \u2192 "${link.href}" target not found in DOM`);
+    }
+  } catch {
+  }
+  return issues;
+}
+async function probeContextMenus(page) {
+  const results = [];
+  try {
+    const targets = await page.$$('canvas, [data-contextmenu], [oncontextmenu], [class*="chart" i]:not(script)');
+    for (const t of targets.slice(0, 4)) {
+      try {
+        await t.click({ button: "right", timeout: 1e3 });
+        await page.waitForTimeout(350);
+        const items = await page.evaluate(
+          () => Array.from(document.querySelectorAll('[role="menu"] [role="menuitem"], .context-menu-item, [class*="contextmenu" i] li')).map((el) => el.innerText?.trim()).filter(Boolean).slice(0, 12)
+        );
+        if (items.length > 0) {
+          const trigger = await t.evaluate((el) => el.getAttribute("aria-label") || el.className.slice(0, 40) || el.tagName).catch(() => "");
+          results.push({ trigger, items });
+        }
+        await page.keyboard.press("Escape").catch(() => {
+        });
+      } catch {
+      }
+    }
+  } catch {
+  }
+  return results;
+}
+async function probeCarousels(page) {
+  const results = [];
+  try {
+    const CAROUSEL_SEL = '[class*="carousel" i], [class*="slider" i], [class*="swiper" i], [role="region"][aria-roledescription*="carousel" i]';
+    const hasCarousel = await page.evaluate((sel) => !!document.querySelector(sel), CAROUSEL_SEL);
+    if (!hasCarousel) return results;
+    const NEXT_SEL = '[aria-label*="next" i], [aria-label*="Next"], [class*="next" i]:not([class*="footnext"]):not(a), button[class*="arrow-right" i], button[class*="chevron-right" i]';
+    const slide0 = await page.evaluate((carSel) => {
+      const car = document.querySelector(carSel);
+      if (!car) return null;
+      const active = car.querySelector('[aria-hidden="false"], .active, [class*="active" i], [class*="current" i]');
+      const target = active ?? car;
+      return { heading: target.querySelector("h2,h3,h4")?.textContent?.trim() ?? "", content: target.innerText?.replace(/\s+/g, " ").trim().slice(0, 300) ?? "" };
+    }, CAROUSEL_SEL);
+    if (slide0?.content && slide0.content.length > 15) results.push({ slideIndex: 0, heading: slide0.heading || void 0, content: slide0.content });
+    const seen = new Set(slide0?.content ? [slide0.content] : []);
+    for (let i = 1; i < 8; i++) {
+      try {
+        const nextBtn = page.locator(NEXT_SEL).first();
+        const visible = await nextBtn.isVisible().catch(() => false);
+        if (!visible) break;
+        await nextBtn.click({ timeout: 1e3 });
+        await page.waitForTimeout(600);
+        const slide = await page.evaluate((carSel) => {
+          const car = document.querySelector(carSel);
+          if (!car) return null;
+          const active = car.querySelector('[aria-hidden="false"], .active, [class*="active" i], [class*="current" i]');
+          const target = active ?? car;
+          return { heading: target.querySelector("h2,h3,h4")?.textContent?.trim() ?? "", content: target.innerText?.replace(/\s+/g, " ").trim().slice(0, 300) ?? "" };
+        }, CAROUSEL_SEL);
+        if (!slide?.content || slide.content.length < 15 || seen.has(slide.content)) break;
+        seen.add(slide.content);
+        results.push({ slideIndex: i, heading: slide.heading || void 0, content: slide.content });
+      } catch {
+        break;
+      }
+    }
+  } catch {
+  }
+  return results;
+}
+async function probeWizardSteps(page) {
+  const results = [];
+  try {
+    const NEXT_LABELS = /^(next|continue|proceed|step \d|›|→|forward)/i;
+    const BACK_LABELS = /^(back|previous|prev|‹|←)/i;
+    const WIZARD_INDICATORS = '[role="progressbar"], [class*="stepper" i], [class*="wizard" i], [class*="progress-step" i], [aria-label*="step" i]';
+    const isWizard = await page.evaluate((sel) => !!document.querySelector(sel), WIZARD_INDICATORS);
+    if (!isWizard) return results;
+    const captureStep = async (stepIndex) => {
+      const data = await page.evaluate(() => {
+        const content = document.querySelector('main, [role="main"], form')?.innerText?.replace(/\s+/g, " ").trim().slice(0, 400) ?? "";
+        const stepLabel = document.querySelector('[aria-current="step"], [class*="step-label" i], [class*="stepper-label" i]')?.textContent?.trim() ?? "";
+        const fields = Array.from(document.querySelectorAll('input:not([type="hidden"]), select, textarea')).map((el) => el.getAttribute("aria-label") || el.placeholder || el.name || "").filter(Boolean).slice(0, 10);
+        return { content, stepLabel, fields };
+      });
+      return { stepIndex, stepLabel: data.stepLabel || void 0, content: data.content, fields: data.fields };
+    };
+    results.push(await captureStep(0));
+    for (let i = 1; i < 6; i++) {
+      try {
+        const nextLabel = await page.evaluate((nextRe) => {
+          const btns = Array.from(document.querySelectorAll('button, [role="button"]'));
+          const btn2 = btns.find((b) => new RegExp(nextRe).test(b.innerText?.trim() || ""));
+          return btn2?.innerText?.trim() || null;
+        }, NEXT_LABELS.source);
+        if (!nextLabel) break;
+        const btn = page.locator(`button:has-text("${nextLabel}")`).first();
+        await btn.click({ timeout: 1500 });
+        await page.waitForTimeout(800);
+        results.push(await captureStep(i));
+      } catch {
+        break;
+      }
+    }
+  } catch {
+  }
+  return results;
+}
+async function detectFocusTraps(page) {
+  const results = [];
+  try {
+    const openDialogs = await page.evaluate(() => {
+      return Array.from(document.querySelectorAll('[role="dialog"], [role="alertdialog"]')).filter((el) => window.getComputedStyle(el).display !== "none").map((el) => el.getAttribute("aria-label") || el.querySelector('[role="heading"]')?.textContent?.trim() || "dialog");
+    });
+    for (const dialogLabel of openDialogs.slice(0, 3)) {
+      const sampled = [];
+      let allInside = true;
+      for (let i = 0; i < 8; i++) {
+        await page.keyboard.press("Tab");
+        await page.waitForTimeout(80);
+        const focused = await page.evaluate(() => {
+          const el = document.activeElement;
+          const inDialog = !!el?.closest('[role="dialog"], [role="alertdialog"]');
+          return { label: el?.getAttribute("aria-label") || el?.textContent?.trim().slice(0, 40) || el?.tagName || "", inDialog };
+        });
+        sampled.push(focused.label);
+        if (!focused.inDialog) {
+          allInside = false;
+          break;
+        }
+      }
+      results.push({ modalTrigger: dialogLabel, focusTrapped: allInside, sampledFocused: sampled });
+    }
+  } catch {
+  }
+  return results;
+}
+async function probeTabPanels(page) {
+  const results = [];
+  try {
+    const tabs = await page.$$('[role="tab"]:not([aria-disabled="true"])');
+    for (const tab of tabs.slice(0, 10)) {
+      try {
+        const tabLabel = await tab.evaluate((el) => el.innerText?.trim() || el.getAttribute("aria-label") || "");
+        if (!tabLabel) continue;
+        await tab.click({ timeout: 1e3 });
+        await page.waitForTimeout(400);
+        const panelContent = await page.evaluate(() => {
+          const panel = document.querySelector('[role="tabpanel"]:not([hidden])');
+          if (!panel) return null;
+          const style = window.getComputedStyle(panel);
+          if (style.display === "none" || style.visibility === "hidden") return null;
+          return {
+            heading: panel.querySelector("h2,h3,h4")?.textContent?.trim() ?? "",
+            content: panel.innerText?.replace(/\s+/g, " ").trim().slice(0, 400) ?? ""
+          };
+        });
+        if (panelContent?.content && panelContent.content.length > 15) {
+          results.push({ tabLabel: tabLabel.slice(0, 60), panelContent: panelContent.content, panelHeading: panelContent.heading || void 0 });
+        }
+      } catch {
+      }
+    }
+  } catch {
+  }
+  return results;
+}
+async function probeTooltips(page) {
+  const results = [];
+  try {
+    const candidates = await page.evaluate(() => {
+      return Array.from(document.querySelectorAll('[title], [aria-describedby], [data-tooltip], [data-tip], [class*="tooltip" i]')).filter((el) => {
+        const style = window.getComputedStyle(el);
+        return style.display !== "none" && style.visibility !== "hidden";
+      }).slice(0, 8).map((el) => ({
+        label: el.innerText?.trim().slice(0, 60) || el.getAttribute("aria-label") || "",
+        title: el.getAttribute("title") || "",
+        describedBy: el.getAttribute("aria-describedby") || "",
+        selector: el.tagName.toLowerCase() + (el.id ? `#${el.id}` : "")
+      }));
+    });
+    for (const c of candidates) {
+      try {
+        if (c.title && c.title.length > 5) {
+          results.push({ triggerLabel: c.label || c.selector, tooltipText: c.title.slice(0, 200) });
+          continue;
+        }
+        if (c.describedBy) {
+          const tooltipText = await page.evaluate((id) => document.getElementById(id)?.textContent?.trim() ?? "", c.describedBy);
+          if (tooltipText && tooltipText.length > 5) {
+            results.push({ triggerLabel: c.label || c.selector, tooltipText: tooltipText.slice(0, 200) });
+          }
+        }
+      } catch {
+      }
+    }
+  } catch {
+  }
+  return results;
+}
 async function probeKeyboardNav(page) {
   const sequence = [];
   const issues = [];